An addition to testing guidelines for COVID-19 could help detect potential super-spreaders, protect healthcare workers and support the healthcare system against the surging onslaught of COVID-19.

Andy Chen

Mar 14, 2020·6 min read

A COVID-19 Dilemma for the Doctor

A doctor has 1 reliable diagnostic test. Two patients show up to the clinic.

1. The patient has a dry cough and fever. A travel history reveals that the patient had returned four days ago from Milan, Italy.
2. The patient has a runny nose, and reports having a mild cold over the past few days, which is mostly resolved. A travel history reveals that the patient had returned four days ago from San Francisco, US.

Who should get the test?

The doctor looks at the official criteria for determining who to test.

OFFICIAL GUIDELINES

From Coronavirus Disease 2019 (COVID-19)
CDNA National Guidelines for Public Health Units

(Communicable Diseases Network Australia)

SUSPECT CASE

A. If the patient satisfies epidemiological and clinical criteria, they are classified as a suspect case.

Epidemiological criteria
* Travel to (including transit through) a country considered to pose a risk of transmission* in the 14 days before the onset of illness.
OR
* Close or casual contact (see Contact definition below) in 14 days before illness onset with a confirmed case of COVID-19.

Clinical criteria
* Fever.
OR
* Acute respiratory infection (e.g. shortness of breath or cough) with or without fever.

…

* Country transmission risk assessment
Higher risk: Mainland China, Iran, Italy, South Korea
Moderate risk: Cambodia, Hong Kong, Indonesia, Japan, Singapore,Thailand

…

Patients meeting the suspect case definition (above) should be tested for SARS-CoV-2.

The doctor sees that:

1. The first patient satisfies clinical criteria (fever and cough), and epidemiological criteria (returned from Italy).
2. The second patient may be ill, but the patient does not satisfy epidemiological criteria (USA is not on the list).

Therefore, patient 1 is deemed a suspect case (likely infection) and receives the diagnostic test. Patient 2 is deemed to not be suspicious enough to warrant testing (unlikely infection).

Another Dilemma for the Doctor

Suppose we have the same two patients. But now, we add new information.

1. The patient has a dry cough and fever. A travel history reveals that the patient had returned four days ago from Milan, Italy. The patient lives with a roommate. As a full-time blog writer, the patient does not have many routine close contacts.
2. The patient has a runny nose, and reports having a mostly-resolved mild cold over the past few days. A travel history reveals that the patient had returned four days ago from San Francisco, USA. The patient lives in a large household of 6 people, studies part-time, works in an aged-care center, and regularly volunteers with local charities.

Who should get the test?

Consider The Idea of “Payoffs”.

There is an idea which is usually more relevant to finance, but can potentially be applied to medicine, and in particular, epidemics.

Two gamblers make a series of 5 predictions, and win or lose money based on their predictions.

Gambler 1
Results: {+1, +2, +1, +1, -10}
Total: -$5
Win %: 4/5 (80% of predictions are correct)

Gambler 2
Results: {-1, -1, -3, -2, +11}
Total: +$4
Win %: 1/5 (20% of predictions are correct)

You can see that despite being more accurate, the first gambler actually performs worse than gambler 2, and the reason for this is because of the skewed payoffs.

Payoff — the amount won or lost as the result of a gamble.

How gambles relate to diagnostic testing

The gamble is analogous to ordering diagnostic tests for patients who may have COVID-19. The doctor predicts whether a patient has COVID-19 or not, and the payoff is the benefit derived from that test result. A negative result is a loss of resources, but a positive doesn’t have to only include treatment of the patient. The benefit of a positive result also includes prevention of a chain of transmissions, which has unbounded potential harm. This is unlike non-contagious diseases which have their potential harm limited to the patient.

The diagnostic guidelines (see above) are designed to increase the probability of a positive result. Test are recommended for the more likely cases, but not the less likely cases. However, these guidelines ignore payoffs — the number of additional transmissions resulting from one case can be very large. It may be desirable to test patients who have many close contacts, even if their symptoms indicate that COVID-19 is unlikely.

An additional criteria for testing, based on payoffs

To decide whether unlikely cases should be tested for COVID-19, it is important to know more about the patient’s life.

• Who do they live with? Parents, children, other relatives?
• Where do they work? Alone, in a small office, or perhaps in a large venue?

If we consider the scenario again.

1. The patient has a dry cough and fever. A travel history reveals that the patient had returned four days ago from Milan, Italy. The patient lives with a roommate, and works as a full-time blog writer.
2. The patient has a runny nose, and reports having a mostly-resolved mild cold over the past few days. A travel history reveals that the patient had returned four days ago from San Francisco, US. The patient lives in a large household of 6 people, studies part-time, works in an aged-care center, and regularly volunteers with local charities.

I now draw two possible social networks to see what the potential for community spread might look like.

Patient 1 has 4 regular contacts while patient 2 has 21 regular contacts.

The current guidelines recommends that the first patient be tested while the second patient not be tested. But with this new way of thinking, the potential harm is too great if patient 2 is infected. Thus, the criteria would also recommend a test for patient 2.

Other Considerations With Implementation

One issue with the “payoff” view is that it would affect equal access to healthcare services.

Assuming that the patient is an unlikely case:

• Homeless people might not be given tests
• Politicians, wealthy businessmen and the “well-connected” (many social connections) would automatically get prioritized for tests.

To lessen these issues, the guidelines could:

• Prioritize those with vulnerable dependents (aged-care workers, caretakers of the disabled)
• Prioritize those who are irreplaceable and have critical roles in fighting the epidemic (health-care workers, emergency workers, senior ministers)
• Other cases which are less suspicious can be advised to self-isolate.

March 16 Edit: This could be better summarized as anyone who cannot self-isolate, because their absence would necessarily lead to the harm of other lives.

Another issue is that this would require more tests at a time when the supply of tests is already too low.

This is a good criticism. Theoretically, if we could successfully detect and contain risky individuals early, then we could avoid a chain of transmission and reduce the number of future tests. Whether this reduction in future tests balances out the additional tests on potential super-spreaders is uncertain, and should be closely studied.

Nonetheless, I think it is worth considering this idea for future epidemics. If successful, this addition to the guidelines would greatly aid the containment of outbreaks.

Any thoughts are appreciated.